Re: EP93xx PIO IDE driver proposal

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Sergei, thanks for reviewing it.

João, all Sergei's points are valid and here are few more ;)

On Tuesday 12 May 2009 18:49:37 Sergei Shtylyov wrote:
> Hello.
> 
> João Ramos wrote:
> 
> >> I'm not sure what you're getting at with "such advices"... :)
> 
> >> We need to cast somewhere anyway so we may as well cast from 'void *' to
> >> 'unsigned int' where needed and not the other way around (or rather from
> >> 'unsigned int' to 'struct ide_timing *') -- which would be an ugly hack
> >> and could cause maintainability/portability problems later.
> 
> >> BTW it seems like a good occasion to add ide_{get,set}_drivedata() 
> >> helpers
> >> (ala existing ide_{get,set}_hwifdata() ones) to <linux/ide.h> and thus 
> >> make
> >> internal IDE API more coherent.
> 
> >> [ Yes, I know that we may get away with s/unsigned int/unsigned long/
> >>   and casting it to 'struct ide_timing *' for now but it always better
> >>   to at least consider more elegant solution first... ]
> 
>     Changing a lot of drivers is more elegant than not? Doubt it.
> 
> >> Thanks,
> >> Bart
> 
> > Here is a proposed patch to fix 'drive_data' field to a 'void *' type.
> > I've also added the 'ide_{get,set}_drivedata' helpers you hinted, and 
> 
> > I've fixed all the host drivers which used the 'drive_data' field, 
> > making them use the new 'ide_{get,set}_drivedata' helpers and casting 
> > them accordingly to both the new 'drive_data' type and the type of value 
> > being stored.
> 
> > I've attached the patch to avoid line-wrapping and deformed output.
> 
> > If you can, please test the patch with the fixed drivers, I can't test 
> > those since I don't have hardware for it.
> > However, I will test for compilation.
> 
> > Please comment, and if there's any fix to be made, just let me know.
> 
> > Regards,
> > João Ramos
> 
> > ------------------------------------------------------------------------
> > 
> > Change ide_drive_t 'drive_data' field from 'unsigned int' type to 'void *' type,
> > allowing a wider range of values/types to be stored in this field.
> > Added 'ide_get_drivedata' and 'ide_set_drivedata' helpers to get and set
> > the 'drive_data' field.
> > Fixed all host drivers to maintain coherency with the change in the 'drive_data'
> > field type.
> > 
> > Signed-off-by: Joao Ramos <joao.ramos@xxxxxxx>
> > Cc: Bartlomiej Zolnierkiewicz <bzolnier@xxxxxxxxx>
> > Cc: Sergei Shtylyov <sshtylyov@xxxxxxxxxxxxxxxxx>
> > Cc: Alan Cox <alan@xxxxxxxxxxxxxxxxxxx>
> > 
> > ---
> > 
> > diff --git a/drivers/ide/cmd64x.c b/drivers/ide/cmd64x.c
> > index 80b777e..c8df34c 100644
> > --- a/drivers/ide/cmd64x.c
> > +++ b/drivers/ide/cmd64x.c
> > @@ -140,10 +140,11 @@ static void cmd64x_tune_pio(ide_drive_t *drive, const u8 pio)
> >  	if (hwif->channel) {
> >  		ide_drive_t *pair = ide_get_pair_dev(drive);
> >  
> > -		drive->drive_data = setup_count;
> > +		ide_set_drivedata(drive, (void *) setup_count);

(void *)setup_count would be a preferred coding style
[this holds for all such constructs in the patch]
  
> >  		if (pair)
> > -			setup_count = max_t(u8, setup_count, pair->drive_data);
> > +			setup_count = max_t(u8, setup_count,
> > +							(u8) ide_get_drivedata(pair));
> 
>     You've over-indented it and the (u8) cast is not needed with max_t() 
> which performs the cast internally.
>     Personally, I prefer this style for the multi-line statement:
> 
> 			setup_count = max_t(u8, setup_count,
> 					    ide_get_drivedata(pair));
> 
> > diff --git a/drivers/ide/cs5536.c b/drivers/ide/cs5536.c
> > index 0332a95..f9e349e 100644
> > --- a/drivers/ide/cs5536.c
> > +++ b/drivers/ide/cs5536.c
> > @@ -143,6 +143,8 @@ static void cs5536_set_pio_mode(ide_drive_t *drive, const u8 pio)
> >  		0x99, 0x92, 0x90, 0x22, 0x20,
> >  	};
> >  
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Call it 'timings' please.

and please move it after 'cshift' variable below

> > +
> >  	struct pci_dev *pdev = to_pci_dev(drive->hwif->dev);
> >  	ide_drive_t *pair = ide_get_pair_dev(drive);
> >  	int cshift = (drive->dn & 1) ? IDE_CAST_D1_SHIFT : IDE_CAST_D0_SHIFT;
> [...]
> > @@ -184,6 +187,8 @@ static void cs5536_set_dma_mode(ide_drive_t *drive, const u8 mode)
> >  		0x67, 0x21, 0x20,
> >  	};
> >  
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     And 'timings' again...
> 
> > +
> >  	struct pci_dev *pdev = to_pci_dev(drive->hwif->dev);
> >  	int dshift = (drive->dn & 1) ? IDE_D1_SHIFT : IDE_D0_SHIFT;
> >  	u32 etc;
> [...]
> > @@ -204,9 +210,11 @@ static void cs5536_set_dma_mode(ide_drive_t *drive, const u8 mode)
> >  
> >  static void cs5536_dma_start(ide_drive_t *drive)
> >  {
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     And again...
> 
> > +
> >  	if (drive->current_speed < XFER_UDMA_0 &&
> > -	    (drive->drive_data >> 8) != (drive->drive_data & IDE_DRV_MASK))
> > -		cs5536_program_dtc(drive, drive->drive_data >> 8);
> > +	    (drivedata >> 8) != (drivedata & IDE_DRV_MASK))
> > +		cs5536_program_dtc(drive, drivedata >> 8);
> >  
> >  	ide_dma_start(drive);
> >  }
> > @@ -214,10 +222,11 @@ static void cs5536_dma_start(ide_drive_t *drive)
> >  static int cs5536_dma_end(ide_drive_t *drive)
> >  {
> >  	int ret = ide_dma_end(drive);
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Here too...
> 
> >  
> >  	if (drive->current_speed < XFER_UDMA_0 &&
> > -	    (drive->drive_data >> 8) != (drive->drive_data & IDE_DRV_MASK))
> > -		cs5536_program_dtc(drive, drive->drive_data & IDE_DRV_MASK);
> > +	    (drivedata >> 8) != (drivedata & IDE_DRV_MASK))
> > +		cs5536_program_dtc(drive, drivedata & IDE_DRV_MASK);
> >  
> >  	return ret;
> >  }
> > diff --git a/drivers/ide/ht6560b.c b/drivers/ide/ht6560b.c
> > index 2fb0f29..158aeae 100644
> > --- a/drivers/ide/ht6560b.c
> > +++ b/drivers/ide/ht6560b.c
> > @@ -44,7 +44,12 @@
> >   *    bit3 (0x08): "1" 3 cycle time, "0" 2 cycle time	      (?)
> >   */
> >  #define HT_CONFIG_PORT	  0x3e6
> > -#define HT_CONFIG(drivea) (u8)(((drivea)->drive_data & 0xff00) >> 8)
> > +
> > +static inline u8 HT_CONFIG(ide_drive_t *drive)
> > +{
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Call it 'config' (or maybe just 'data' or 'info') and insert an empty 
> line after declaration please.

Actually this one can be done without 'drivedata' variable.

> > +	return (drivedata & 0xff00) >> 8;
> > +}
> >  /*
> >   * FIFO + PREFETCH (both a/b-model)
> >   */
> > @@ -90,7 +95,12 @@
> >   * Active Time for each drive. Smaller value gives higher speed.
> >   * In case of failures you should probably fall back to a higher value.
> >   */
> > -#define HT_TIMING(drivea) (u8)((drivea)->drive_data & 0x00ff)
> > +static inline u8 HT_TIMING(ide_drive_t *drive)
> > +{
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Same here.

this one also

> > +	return drivedata & 0x00ff;
> > +}
> > +
> >  #define HT_TIMING_DEFAULT 0xff
> >  
> >  /*
> > @@ -244,6 +254,7 @@ static void ht_set_prefetch(ide_drive_t *drive, u8 state)
> >  {
> >  	unsigned long flags;
> >  	int t = HT_PREFETCH_MODE << 8;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Call 'config' (or maybe just 'data' or 'info') please.
> 
> > @@ -270,6 +283,7 @@ static void ht6560b_set_pio_mode(ide_drive_t *drive, const u8 pio)
> >  {
> >  	unsigned long flags;
> >  	u8 timing;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Same here...
> 
> [...]
> 
> > @@ -299,7 +314,7 @@ static void __init ht6560b_init_dev(ide_drive_t *drive)
> >  	if (hwif->channel)
> >  		t |= (HT_SECONDARY_IF << 8);
> >  
> > -	drive->drive_data = t;
> > +	ide_set_drivedata(drive, (void *)t);
> 
>     Oh, they call it just 't'! Nice name too. :-D
> 
> > diff --git a/drivers/ide/icside.c b/drivers/ide/icside.c
> > index 4e16ce6..c96f87a 100644
> > --- a/drivers/ide/icside.c
> > +++ b/drivers/ide/icside.c
> > @@ -267,10 +267,11 @@ static void icside_set_dma_mode(ide_drive_t *drive, const u8 xfer_mode)
> >  	if (use_dma_info && drive->id[ATA_ID_EIDE_DMA_TIME] > cycle_time)
> >  		cycle_time = drive->id[ATA_ID_EIDE_DMA_TIME];
> >  
> > -	drive->drive_data = cycle_time;
> > +	ide_set_drivedata(drive, (void *)cycle_time);
> >  
> >  	printk("%s: %s selected (peak %dMB/s)\n", drive->name,
> > -		ide_xfer_verbose(xfer_mode), 2000 / drive->drive_data);
> > +		ide_xfer_verbose(xfer_mode),
> > +		2000 / (int) ide_get_drivedata(drive));
> >  }
> 
>     I see that you're not consistent about that space after the typeacst... 
> either put it there or not, not both. :-)
> 
> > diff --git a/drivers/ide/opti621.c b/drivers/ide/opti621.c
> > index 6048eda..d545f01 100644
> > --- a/drivers/ide/opti621.c
> > +++ b/drivers/ide/opti621.c
> > @@ -139,6 +139,8 @@ static void opti621_set_pio_mode(ide_drive_t *drive, const u8 pio)
> >  	ide_drive_t *pair = ide_get_pair_dev(drive);
> >  	unsigned long flags;
> >  	u8 tim, misc, addr_pio = pio, clk;
> > +	int drivedata = XFER_PIO_0 + pio;
> 
>     Please name it 'mode'...
> 
> > +	int pair_drivedata;
> 
>     ... and 'pair_mode'.
> 
> > @@ -150,11 +152,12 @@ static void opti621_set_pio_mode(ide_drive_t *drive, const u8 pio)
> >  		{ 0x48, 0x34, 0x21, 0x10, 0x10 }	/* 25 MHz */
> >  	};
> >  
> > -	drive->drive_data = XFER_PIO_0 + pio;
> > +	ide_set_drivedata(drive, (void *) drivedata);
> >  
> >  	if (pair) {
> > -		if (pair->drive_data && pair->drive_data < drive->drive_data)
> > -			addr_pio = pair->drive_data - XFER_PIO_0;
> > +		pair_drivedata = (int) ide_get_drivedata(pair);
> > +		if (pair_drivedata && pair_drivedata < drivedata)
> > +			addr_pio = pair_drivedata - XFER_PIO_0;
> >  	}
> >  
> >  	spin_lock_irqsave(&opti621_lock, flags);
> > diff --git a/drivers/ide/qd65xx.c b/drivers/ide/qd65xx.c
> > index c9a1349..2cd350c 100644
> > --- a/drivers/ide/qd65xx.c
> > +++ b/drivers/ide/qd65xx.c
> > @@ -180,8 +180,11 @@ static int qd_find_disk_type (ide_drive_t *drive,
> >  
> >  static void qd_set_timing (ide_drive_t *drive, u8 timing)
> >  {
> > -	drive->drive_data &= 0xff00;
> > -	drive->drive_data |= timing;
> > +	int drivedata = (int) ide_get_drivedata(drive);
> 
>     Well, please call it just 'data' (or even 't') in this case.
> 
> > +
> > +	drivedata &= 0xff00;
> > +	drivedata |= timing;
> > +	ide_set_drivedata(drive, (void *)drivedata);
> 
>     Oh, inconsistent spacing again! :-)

also:

@@ -291,8 +294,9 @@ static void __init qd6500_init_dev(ide_drive_t *drive)
        ide_hwif_t *hwif = drive->hwif;
        u8 base = (hwif->config_data & 0xff00) >> 8;
        u8 config = QD_CONFIG(hwif);
+       int drivedata = QD6500_DEF_DATA;
 
-       drive->drive_data = QD6500_DEF_DATA;
+       ide_set_drivedata(drive, (void *) drivedata);

No need for drivedata variable.

 static void __init qd6580_init_dev(ide_drive_t *drive)

@@ -301,6 +305,7 @@ static void __init qd6580_init_dev(ide_drive_t *drive)
        u16 t1, t2;
        u8 base = (hwif->config_data & 0xff00) >> 8;
        u8 config = QD_CONFIG(hwif);
+       int drivedata = (drive->dn & 1) ? t2 : t1;

dito, plus t2 and t1 are not initialized yet
 
        if (hwif->host_flags & IDE_HFLAG_SINGLE) {
                t1 = QD6580_DEF_DATA;
@@ -308,7 +313,7 @@ static void __init qd6580_init_dev(ide_drive_t *drive)
        } else
                t2 = t1 = hwif->channel ? QD6580_DEF_DATA2 : QD6580_DEF_DATA;
 
-       drive->drive_data = (drive->dn & 1) ? t2 : t1;
+       ide_set_drivedata(drive, (void *) drivedata);
 }

> > diff --git a/drivers/ide/sl82c105.c b/drivers/ide/sl82c105.c
> > index b0a4606..5e42eaa 100644
> > --- a/drivers/ide/sl82c105.c
> > +++ b/drivers/ide/sl82c105.c
> > @@ -76,6 +76,7 @@ static void sl82c105_set_pio_mode(ide_drive_t *drive, const u8 pio)
> >  	struct pci_dev *dev	= to_pci_dev(drive->hwif->dev);
> >  	int reg			= 0x44 + drive->dn * 4;
> >  	u16 drv_ctrl;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Please call it 'timings'.
> 
> > @@ -101,6 +103,7 @@ static void sl82c105_set_dma_mode(ide_drive_t *drive, const u8 speed)
> >  {
> >  	static u16 mwdma_timings[] = {0x0707, 0x0201, 0x0200};
> >  	u16 drv_ctrl;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     Here too...
> 
> > @@ -181,10 +185,11 @@ static void sl82c105_dma_start(ide_drive_t *drive)
> >  	ide_hwif_t *hwif	= drive->hwif;
> >  	struct pci_dev *dev	= to_pci_dev(hwif->dev);
> >  	int reg 		= 0x44 + drive->dn * 4;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     And here...
> 
> > @@ -204,12 +209,13 @@ static int sl82c105_dma_end(ide_drive_t *drive)
> >  	struct pci_dev *dev	= to_pci_dev(drive->hwif->dev);
> >  	int reg 		= 0x44 + drive->dn * 4;
> >  	int ret;
> > +	unsigned int drivedata = (unsigned int) ide_get_drivedata(drive);
> 
>     And here...
> 
> > diff --git a/include/linux/ide.h b/include/linux/ide.h
> > index ff65fff..4ab9859 100644
> > --- a/include/linux/ide.h
> > +++ b/include/linux/ide.h
> > @@ -565,7 +565,7 @@ struct ide_drive_s {
> >  
> >  	unsigned int	bios_cyl;	/* BIOS/fdisk/LILO number of cyls */
> >  	unsigned int	cyl;		/* "real" number of cyls */
> > -	unsigned int	drive_data;	/* used by set_pio_mode/dev_select() */
> > +	void	*drive_data;		/* used by set_pio_mode/dev_select() */
> 
>     Please don't break the alignment of the 2nd column.
> 
> >  	unsigned int	failures;	/* current failure count */
> >  	unsigned int	max_failures;	/* maximum allowed failure count */
> >  	u64		probed_capacity;/* initial reported media capacity (ide-cd only currently) */
> > @@ -1570,6 +1570,16 @@ static inline ide_drive_t *ide_get_pair_dev(ide_drive_t *drive)
> >  	return (peer->dev_flags & IDE_DFLAG_PRESENT) ? peer : NULL;
> >  }
> >  
> > +static inline void *ide_get_drivedata(ide_drive_t *drive)
> > +{
> > +	return drive->drive_data;
> > +}
> > +
> > +static inline void ide_set_drivedata(ide_drive_t *drive, void *data)
> > +{
> > +	drive->drive_data = data;
> > +}
> > +
> >  #define ide_port_for_each_dev(i, dev, port) \
> >  	for ((i) = 0; ((dev) = (port)->devices[i]) || (i) < MAX_DRIVES; (i)++)
> >  
> 
>     And don't tell me this patch is "elegant"... :-/

Well, it is makes code more coherent and more maintainable...

Moreover those inline helpers serve as a part of documentation (though
real DocBook documentation would also be nice) for host drivers authors.

Either this or we should remove ide_{get,set}_hwifdata() and allow
host drivers to poke directly also at hwif->hwif_data...
--
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [Linux Filesystems]     [Linux SCSI]     [Linux RAID]     [Git]     [Kernel Newbies]     [Linux Newbie]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Samba]     [Device Mapper]

  Powered by Linux